3 ms·Is that not just traditional OCR applied on top of LLM?by sushid 2y agoIs that not just traditional OCR applied on top of LLM?maxlamb 2y agoNo it’s not, it’s a multimodal transformer model.energy123 2y agoIt's possible they have a software layer that does that. But I was assuming they don't, because the open source multimodal models don't.